Published on February 6, 2025 11:33 PM GMT
Summary
AI may produce a windfall surge in government revenues in 5 to 10 years.I want governments to spending a small fraction of that windfall onretroactively rewarding entities in proportion to how they havecontributed to medical advances, measured by lives saved and sufferingavoided.
Motivations
This post was inspired in part byCritch:
Healthcare, agriculture, food science, education, entertainment, andenvironmental restoration are all important industries that servehumans but not machines. These are industries I want to sustain andadvance, in order to keep the economy caring for humans, and to avoidsuccessionism and industrial dehumanization.
I want to mitigate panic over the coming mass retirement. For now, mostpeople imagine that AI-induced unemployment will be a catastrophe. Someof that is because they can't see how to pay for medical care in aworld where most people depend on welfare. Medicaid and Medicare lookunsustainable in a world where more people depend on those systems, andthe cost of medical care keeps rising. That leads to voters beingsomewhat panicky about AI in a way that fails to focus their attentionon the biggest risks of AI (i.e. world domination by a few AIs or by afew AI-wielding people).
A final motivation is concern that medical progress is too slow tohandle my age-related health risks. I'll highlight two contributingproblems which I expect my proposal will mitigate:
- There's as much reward for marginally beneficial treatments as forfull cures. Since the former seem easier to find, companies havelargely ignored the most valuable goals.Medical research is too dependent on patents for funding. There arelikely a number of cheap, effective treatments that are beingoverlooked because either they're unpatentable, or their patentshave expired.
2025 is a unique time for this proposal: a small fraction of investorsare seeking big AI-related profits. If they're right, we can expect toafford much more medical research now than is currently happening. Theseinvestors forecast a dramatic increase in productivity, which seems likeit would generate unprecedented economic growth, presumably enabling adramatic increase in government spending. There are enough suchinvestors to make a real difference in medical progress. I think it'spossible that my proposal could redirect several billion dollars toresearch that is valuable and neglected.
Designing the Prizes
I propose allocating 1% of windfall US government revenues between nowand 2040 toward these prizes. By windfall, I mean revenues that areclearly in excess of what would be expected if future growth is similarto historical growth. E.g. if revenue growth accelerates to 15%/yearstarting in 2028, revenue in 2032 would be around $16.5 trillion,versus $6.2 trillion expected under normal growth. I'd classify $6trillion of that as a windfall, for a prize fund of $60 billion thatyear. The appendix describes the calculation in more detail.
Who gets these prizes? I want a panel of 3 to 5 of the most respectedAIs to decide. In any given windfall year, I want them to estimate whois responsible for medical advances, and divide the prize fund equallyper QALY saved.
I want to leave most of the details to future AIs, who I expect will bewiser and more trustworthy than current institutions.
Benefits
The prizes would cause some investor money to be redirected towardensuring that many people are healthy.
My proposal would reduce the risk of gradualdisempowermentof the average person.
We're heading toward a world where most institutions may end up focusedalmost entirely on serving machines. Many of those institutions willmeasure their success by criteria such as profits or donations. Theseprizes will ensure that some institutions measure at least some of theirsuccess by a criterion that is more closely connected to human welfare.
My guess is that gradual disempowerment will be not prohibitively hardto avoid once we have AIs that are modestly smarter than the smartesthumans to advise us on how to coordinate. But I'm uncertain enoughabout that that I want to add in whatever precautions we can implementnow.
My proposal would mitigate retirement fears by significantly increasingresearch into treatments that cheaply prevent cancer, cardiovasculardisease, etc. That's a step toward ensuring that programs such asMedicaid won't collapse due to prohibitive costs. The offer of prizeswouldn't be immediately reassuring to the large number of people whodoubt that economic growth will accelerate. However, I expect that bythe time AI causes noticeable unemployment, they'd see some shift incorporate strategies toward cheaper and more effective treatments.
My proposal would divert a modest amount of investment away from ageneral AI capabilities race, providing modestly more time for safetyresearch. This effect will be less than I hope for, but something isbetter than nothing.
Examples
Most of my hopes for better medicine come from the discovery oftreatments that I haven't yet imagined. But I'll try to whet yourappetite by suggesting some examples of currently known treatments thatmight be used more widely as a result of the improved incentives fromthis prize fund. Remember that I'm just suggesting possibilities. I'mnot willing to bet heavily on the value of any specific example.
Healthy Meals: A wide variety of age-related diseases are influencedby diet, most notably diabetes.
I imagine a company that sells convenient healthy prepared meals atbelow cost. Current institutions aren't good enough to compensate thecompany for the health benefits, but it seems likely that transformativeAIs will be wise enough to handle this fairly well.
Affordable Weight Loss: GLP-1 drugs appear to have widespreadhealthbenefits,yet lots of people who might benefit from one are unable to obtain one.The obstacles typically boil down tocost.
Conceivably the prizes could persuade pharma companies to reduce GLP-1drug prices, in order to get prize money by getting the drugs to morepatients. But I don't expect pharma companies to take AI seriouslyenough for that effect to be significant.
AI-pilled investors could subsidize prescriptions. I don't have a clearintuition as to how well that would work.
A somewhat more promising strategy involves Modere'sCurb(a supplement that influences GLP-1). A startup funded by AI-pilled VCscould buy Modere, pump lots of money into advertising Curb, and cutCurb's price a bit. They'd lose moderate amounts of money for the nextfew years, for what they'd see as more than a 10% chance of getting $1billion in windfall prize money.
Malaria Vaccine: A malaria vaccine recently became available. As faras I can tell, it is not being rolledoutas fast as it would be if the world treated malaria deaths as a crisis.It seems likely that applying more money to the process could cheaplysave lives here.
The malaria example assumes that the prizes would be awarded based onglobal lives saved, but I don't intend to express a strong opinion asto whether we should aim for that, versus a more politically feasiblenational program.
I expect that exercise and mental health therapy are two other areaswhere there's big room for improvement if the incentives becomesufficiently good. I don't have clear enough ideas there to producerealistic proposals.
As you can see from these examples, I'm not sticking to a traditionalnotion of medicine. I want the prize committee to focus on QALY'ssaved, and use a broad definition of healthcare when deciding whattreatments to reward.
Problems
A promise from Trump (as with Biden) is not, by itself, worth the tweetit's written on.
A law passed by congress is worth a bit more, but still not as much as Iwould like.
I see some risk that the proposal will cause AI companies to insertbiases into their AIs about how the prizes should be allocated. Thisseems not much of a problem at the early stages of implementation, whencompanies that are competing to get prizes will have trouble predictingthe characteristics of the AIs that will judge the prizes. It willbecome a bit more of a concern later on. Maybe there needs to be somerandomness in which AIs get selected for the panel?
Conclusion
I've designed this proposal so that there will be minimal opposition tocongress enacting it into law: it looks costless under the normalscenarios that most people use to evaluate such proposals, and veryaffordable under more optimistic scenarios. The downside is thatbenefits also look fairly unlikely under scenarios that assume normalcy,so approximately zero politicians are likely to devote much thought topromoting it.
Weaker versions of this proposal are possible. E.g. Bill Gates couldpromise 10% of the windfall profits from his foundation's investments.
Appendix
Here's an outline of how I calculated the prize fund:
ExpectedRevenue(Y) = BaselineRevenue × (1 + B)^Y * (1 + σ)
Windfall(Year Y) = ActualRevenue(Y) - ExpectedRevenue(Y)
RewardableWindfall(Y) = max(0, Windfall(Y) - ThresholdPercentage ×ExpectedRevenue(Y))
Where: - BaselineRevenue = Tax revenue from the year before policyimplementation - B = Historical baseline growth rate - σ = Additionalbuffer (perhaps 2 standard deviations of historical growth variation) -Y = Years since policy implementation
I neglected inflation. An actual law should adjust for inflation.
I've put some code that I used in a comment on myblog.
What time period should the prize fund from any given year apply to?
Let's imagine an example case: The law was enacted on January 1, 2025.In 2031, the prize fund is $10 billion. In 2032, the prize fund is $50billion.
I suggest applying the $10 billion solely to reward actions made in2025 that save QALYs compared to the baseline. Then in 2032, apply the$50 billion to reward actions from both 2025 and 2026. And so on, up toa total of 5 historical years being rewarded each year.
There should probably be some provision for altering the prize fundafter, say, 2040, since it will become less appropriate to use forecastsfrom 2025 to classify revenues as a windfall. Maybe set it to somestable value that's modestly lower than the 2040 prize fund?
I'm tempted to offer prize money to legislators and FDA employees whohelp with these advances, but I'm concerned that conflict of interestrules would make that inappropriate.
Discuss
